Finding high-frequency customers (use cases)

Oji Udezue
4 min readNov 30, 2022

A summary of how to find your best customers as quickly as is humanly possible

Since I wrote the diagnosing a unicorn framework, motivated founders of all stripes have been hunting me down to ask a set of related questions. They boil down to:

  • How do I find the customers who are using my software in a high-frequency manner?
  • How do I pick a set of customers to get a beachhead?

Most founders want to build large successful horizontal companies where ‘everyone’ loves their use case. The advice I have given in the unicorn framework is anti-that (at least initially), so it makes sense that founders want to know what to do with the advice.

I recently sat down with a Bridgewater alum building software that helps workers manage context switching. This is a pernicious problem at work (I’m looking at you Slack) and a fairly common startup inspiration because ALL of US feel this tax on our attention and time. But an elegant solution that people will buy is not at all obvious in this problem space, because it can be hard to achieve a threshold of benefit.

One key principle I share with founders is the idea of a threshold of benefit — that your software solution has to solve people’s problems by a factor of +3x productivity gains. Or +3x augmentation (i.e. giving a customer superpowers). Compressing a worker’s workflow by just 2x is in the zone of benefit where they have to weigh the relative ROI of changing their habits, existing methods, and tools. Inertia is quite a powerful force for human behavior and workflows. At 3x workflow compression, people start to notice because the reward of change is high enough to invest in it. All software demand change of some sort (even if it’s just acquisition costs), and therefore the idea of a threshold of benefit is critical to delivering value that is important to exchange economic value for.

One key hack to solving a threshold problem over time is that different customers perceive benefits differently. So you should start with customers who perceive/achieve the highest benefit with your solution. These are usually your best ideal customer profiles (ICP) and early adopters. So how do you find and isolate those customers?

  1. First, if you’re doing it right, you identify and build software for these people from jump i.e. you identify their general use case and pain with deep knowledge (customer discovery or extended personal experience) and build specifically for that ICP.
  2. Since most people don’t necessarily do this or have a more general hypothesis, the next best thing is to expose (release) your software early to as many different types of target customers as you can. (This is easier said than done because startups often don’t have the resources to do wide releases and are usually talking to a narrow set of proto-customers to facilitate building, in the first place). Nevertheless, you have to find a way to vary the kind of person who gets access to your early release, unless you have done #1 really really well.).
  3. Thirdly, you observe (product analytics, surveys) to see the most successful and activated customers who use your software much more than others, and then do some research into the use cases they are addressing/adopting. Those use cases are your clue to who finds your software compelling and whose use cases will likely be high frequency.

A few observations:

  • Finding and honing an ICP earlier through customer discovery is much more efficient and low-cost than later. Usually by an order of magnitude.
  • However, ICP definition is like an onion, it should be a subject of constant refinement. It can be hard to get to the center without a software release. Put another way, until you have customers excitedly using your company’s software, a bunch of the pre-build research is theoretical.
  • Resist the urge to keep the exposure to your software small and then only to ‘warm customers’ — people you know and have talked to, who are sympathetic to your cause. Don’t hunt for ‘good reviews’ unless as part of a marketing campaign. There are usually more customers in your future than in your past as a successful software company. The most important thing is learning quickly and this absolutely requires as many people ‘testing’ your software, including those who decide it’s not ready and churn*. Don’t fear learning-related churn. It works itself out if you’re learning and improving quickly. The alternative is slow or no learning.

Questions? Hit me up or comment below.

*Don’t put out crap but absolute perfection is not the goal either. Get feedback early and often, even with a small group of real customers.



Oji Udezue

Decent human being. Proud African. Proud American. VP of Product at Follow me: @ojiudezue